cgd-single-middle-layer-optimization

OUT premise — summaries/2026/08/24/shen-2023-icl-not-gd-s8-icl-demonstrations.md

Created 2026-08-25T02:58:33+00:00

The cGD simulation in Shen et al. (2023) optimizes one random middle layer of LLaMA as a partial fine-tune, not full fine-tuning.